Kusuma, Ida Bagus Putu Dharma Kusuma (2025) Sistem Visualisasi Magnetic Resonance Imaging (MRI) Hippocampus Berbasis Platform Web. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5024211068-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5024211068-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy |
Abstract
Segmentasi citra Magnetic Resonance Imaging (MRI) merupakan tahap krusial dalam diagnosis berbagai gangguan neurologis, terutama penyakit Alzheimer. Penelitian ini merancang dan mengimplementasikan sistem visualisasi berbasis web untuk melakukan segmentasi otomatis pada area hippocampus otak dari citra MRI. Proses segmentasi memanfaatkan model deep learning berbasis U-Net yang telah dilatih menggunakan dataset Alzheimer’s Disease Neuroimaging Initiative (ADNI), dengan fokus pada data berformat NIfTI (.nii.gz).Sistem ini dibangun menggunakan arsitektur modern berbasis Express.js pada sisi backend dan Next.js (Re-act) untuk frontend. Backend menyediakan RESTful API untuk manajemen pengguna, unggah citra, serta proses segmentasi, sedangkan frontend menampilkan antarmuka interaktif dengan dukungan Niivue sebagai viewer MRI 3D. Fitur sistem mencakup registrasi pengguna, autentikasi token, unggah dan segmentasi citra, serta visualisasi hasil secara overlay. Pengujian dilakukan melalui evaluasi performa model menggunakan metrik Dice Similarity Coefficient sebesar 0.906, pengujian API dengan Postman, uji kompatibilitas lintas browser, serta user testing dengan 10 responden dari latar belakang medis dan teknis. Hasil menunjukkan bahwa sistem ini mampu memberikan hasil segmentasi yang akurat, cepat, dan dapat diakses dengan mudah melalui browser, serta potensial diterapkan sebagai alat bantu diagnosis modern.
================================================================================================================================
Magnetic Resonance Imaging (MRI) segmentation is a critical task in the diagnosis of neurological disorders such as Alzheimer’s disease. This project presents the development of a web-based visualization system that performs automatic segmentation of the hippocampus region from brain MRI images. The segmentation pipeline uses a U-Net-based deep learning model trained on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) datasets, focusingon 3D NIfTI volumes. The system architecture comprises an Express.js backend that handles RESTful API endpoints for user management, image upload, and model inference, and a frontend built with Next.js (React) that integrates Niivue for real-time 3D MRI visualization. Core features include user authentication, upload of medical images, server-side segmentation, and overlay display of the result. Testing involved model evaluation with a Dice Similarity Coef- ficient of 0.906, functional API testing using Postman, cross-browser compatibility validation, and usability testing with 10 respondents from healthcare and tech fields. Results indicate the system provides accurate, efficient, and accessible segmentation directly through a browser. This platform shows strong potential as a supportive tool for medical professionals in clinical diagnostics.
Actions (login required)
 |
View Item |